How does R handle high-concurrency API serving architecture using Plumber and load balancers?

R APIs scale using stateless Plumber services behind load balancers with horizontal scaling.

Plumber itself is single-process per instance, so high concurrency is achieved by running multiple instances behind Nginx, Kubernetes Ingress, or cloud load balancers. Requests are distributed across replicas. Each instance must remain stateless, pushing session/state to Redis, databases, or object storage.